Giannis Daras
Postdoctoral Associate
Austin, Texas, United States
Summary
👤
Senior🎓
Top SchoolGiannis Daras is a Postdoctoral Associate at MIT working with Costis Daskalakis and Antonio Torralba, focusing on generative models that learn from limited, corrupted, or out-of-distribution data. With a PhD from UT Austin under Alexandros G. Dimakis, he blends theoretical insights with practical ML tooling to address real-world data challenges. He has contributed to open-source ML infrastructure, notably implementing and refactoring a PyTorch-based multiheaded attention layer in the thinc library and adding visualization capabilities that aid model interpretability. His industry experience includes a Google Research internship and a stint as a Machine Learning Researcher at Explosion, where he worked on transformer architectures for spaCy, and he co-founded Ratle, a fintech startup. He is deeply committed to applying generative modeling to scientific domains, and he brings a rare combination of academic rigor, hands-on software engineering, and startup-building experience. Based in Austin, Texas, he spans the academic and open-source ecosystems, actively shaping neural architectures and their practical deployment.
10 years of coding experience
3 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at The University of Texas at AustinHigh School, 19.7/20, High School, 19.7/20 at Model Experimental Lyceum of AnavrytaHigh School, 19.7/20, High School, 19.7/20 at Model Experimental Gymnasium of AnavrytaEngineer’s Degree, Electrical and Computer Engineering, Engineer’s Degree, Electrical and Computer Engineering at National Technical University of Athens
English, German, Greek
Github Skills (10)
pytorch10
machine-learning10
deep-learning10
python10
nlp9
artificial-intelligence8
functional-programming7
tensorflow5
tensorflow25
jax4
Programming languages (6)
Github contributions (5)
explosion/thinc
Apr 2019 - Jun 2019
🔮 A refreshing functional take on deep learning, compatible with your favorite libraries
Role in this project:
ML Engineer
ML EngineerContributions:18 commits, 1 PR in 1 month
Contributions summary:Giannis primarily focused on implementing and refactoring a PyTorch-based multiheaded attention layer within the `thinc` library. Their work involved integrating a tested implementation of the PytorchMultiHeadedAttention module, adding attention visualization capabilities, and refactoring existing attention mechanisms. The user's commits also included modifications to support the visualization features within the attention layer and refactoring of the attention code.
pythonfunctional-programmingtensorflowtype-checkingnatural-language-processing
giannisdaras/thinc
Dec 2018 - May 2019
🔮 spaCy's Machine Learning library for NLP in Python
Contributions:304 commits, 1 PR, 221 pushes in 5 months
nlppythonmachine-learningspacymachine-learning-library
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.